Accelerating the next technology revolution EUVL Optics Contamination from Resist Outgassing; Status Overview Kevin Orvek 1, Greg Denbeaux 2, Alin Antohe 2, Rashi Garg 2, Chimaobi Mbanaso 2 1 SEMATECH 2 CNSE Copyright 2008 SEMATECH, Inc. SEMATECH, and the SEMATECH logo are registered servicemarks of SEMATECH, Inc. International SEMATECH Manufacturing Initiative, ISMI, Advanced Materials Research Center and AMRC are servicemarks of SEMATECH, Inc. All other servicemarks and trademarks are the property of their respective owners.
INTRODUCTION For five years, the semiconductor industry has been testing EUVL resist outgassing levels out of concern for potential contamination of exposure tool optics. Many techniques have been used to measure resist outgassing, and early correlation tests showed widely varying values from the different tests and sites. The industry is now re-evaluating the resist outgas issue with the advent of full-field tools coming on line and the rapidly increasing number of resist samples that will be required to accelerate EUVL resist development. The key issues are whether Resist outgassing is a primary contributor to optics contamination Resist outgas testing is providing us any real information Resist outgas testing should be continued 5 November 2008 2
HISTORY EUV Optics Contamination and Resist Outgassing 2002 - Carbon contamination reported in illuminator of EUV Engineering Test Stand at Sandia Laboratory [1] - attributed to trace organics in system. 2003 - Intel Corporation, concerned about photoresist, established a test methodology for their EUV micro-exposure tool (MET) [2] Specification: hydrocarbon outgas [CxHy] 6.5E+13 molecules/cm 2 Later adopted by SEMATECH for LBNL, Albany MET tools 2004 - ASML preliminary spec [CxHy] 4.7 E+13 molecules/(cm 2 sec) [3] 2006 - Nikon preliminary spec [CxHy] 7.0 E+12 molecules/(cm 2 sec) [4] 2006 - SEMATECH/Intel raises spec to [CxHy] 6.5E+14 molecules/cm 2 2006 - ASML adopts new test procedure based upon witness damage [5] Spec allows < 2.0% reflectivity loss to witness sample near resist 2007 - SEMATECH [6] and Intel Corporation [7] report carbon contamination in the illuminator optics in the MET tools 5 November 2008 3
Overview of Two Approaches to Resist Outgas Measurement Two approaches are currently used to set specifications on resist outgassing: 1. Count the number of outgas molecules and set a specified limit. e.g., SEMATECH spec of total hydrocarbons based on RGA counting Can be reported as total molecules or as an outgas rate 2. Determine how much damage (reflectivity loss) the resist contributes to a witness sample and set an amount that is permissible. e.g., ASML witness plate testing In the end, both approaches should be based upon some reasoning of how much outgassing is allowed in order to meet throughput and optics lifetime requirements in EUVL scanners. Should take into account in situ protection and/or in situ cleaning techniques if they are available on the scanner 5 November 2008 4
Counting Molecules Approach: How Intel/SEMATECH Specification was Determined MET primary mirror M2 assumed to be the optic at greatest risk from resist outgas products (only optical lens surface with a direct line-of-sight to the resist). 2003/2004 /2005 Spec M2s permitted to acquire 1 monolayer of carbon after 3 years (about 0.3% reflectivity loss) Usage: 64 fields/wafer, 30 wafers/week, 48 weeks/year, 3 years. Surface impingement: 100% of all outgas molecules hit M2. Surface sticking probability: 100% of all molecules that hit stick to M2. Molecular disassociation: Each molecule that sticks generates 10 carbon atoms. 2006 current Spec Same 64 fields/wafer, 36 wafers/week, 40 weeks/year, 3 years. 20% of all outgas molecules hit M2. 85% of all molecules that hit stick to M2. Each molecule that sticks generates 5 carbon atoms. 6.5E+13 molecules/cm 2 6.5E+14 molecules/cm 2 5 November 2008 5
SEMATECH/Intel Approach Still very conservative assumptions in current resist outgas spec level. Yet most resists tested in 2007 passed SEMATECH/Intel outgassing limit. outgas rate - molecules/(cm2sec) 4.5E+14 4.0E+14 3.5E+14 3.0E+14 2.5E+14 2.0E+14 1.5E+14 1.0E+14 5.0E+13 Outgas Rate @ 10mW/cm2 SEMATECH spec converted to rate based upon 10 mw/cm 2 power level of Alpha/Beta tools SEMATECH 2007 spec ASML original spec 0.0E+00 51 49 47 45 43 41 39 37 35 33 31 29 27 25 23 21 19 17 15 13 11 9 7 5 3 1 resist number Nikon original spec 119 tested out of 224+ used on MET tools in 2007, 51 commercial shown Sometimes we allowed resists on MET tools even if they failed our spec. How have we done on protecting the MET tools? 5 November 2008 6
Optics Contamination on SEMATECH s MET Tool SEMATECH has swapped illuminator mirror optics twice due to contamination With each swap, total system power has been restored to original value Intel has experienced same response No sign of significant accumulation of carbon on MET primary optics 400 350 300 shots/mj 250 200 150 100 50 New illuminator optics installed 0 0 200 400 600 800 1000 1200 1400 accumulated dose (J) at sensor Shots required per 1 mj of dose as a function of accumulated dose It looks like Sandia was correct in 2002; cause is residual organics 5 November 2008 7
Determining Damage Approach: ASML s Witness Plate Test Methodology Witness Mirror M1 Resist on tape, exposed area ~ ten 300 mm wafers N. Harned, Mirror Reflectivity Loss and Resist Outgassing Rates, IEUVI Optics Contamination TWG, February 28, 2008 Witness mirror M1 allowed to have reflectivity loss just below 2%!!! That s a huge amount of carbon allowed (~2 nm) due to ASML outgas mitigation capability on ADT tools. Can we get that much carbon from resist outgassing? 5 November 2008 8
How Much Contamination is Due to Resist? Our witness plate testing shows contamination due to resist is swamped by contamination due to the chamber Residual Organics Again!! Pre-test, chamber no resist Test with resist Post-test, chamber no resist Average chamber contribution Resist contribution (subtract chamber) % Reflectivity Change -0.8% -1.3% -1.1% - 0.95% - 0.35% What if we swamp the chamber with typical resist outgas species? Directly inject at high concentrations of ~ 1x10-6 Torr Benzene Tert-butanol Diphenyl Sulfide (sulfur compound) 5 November 2008 9
Contamination Studies of Injected Resist Outgas Species - continued Chosen species for injection and exposure of mirrors to measure contamination Sulfur containing feared from 193 nm days 5 November 2008 10
Contamination Studies of Injected Resist Outgas Species - continued Chamber conditions Clean Chamber (background) Reflectivity results due to contamination from these species Chamber pressure (Torr) Exposure time (hours) Total dose (J/cm 2 ) No significant reflectivity loss for these species at these pressures & doses All injected species R/R Chamber Background Contribution! Residual Organics Again!! Number of pulses (millions) Reflectivity change (ΔR/R) % 2.5 x 10-8 8 29 36-0.35 Benzene 1 x 10-6 8 29 36-0.35 Tert-Butanol 3 x 10-6 8 11.5 36 + 0.09 Diphenyl Sulfide 1 x 10-6 4.2 15 19-0.1 Diphenyl Sulfide 1 x 10-6 3.6 13 16 + 0.23 Diphenyl Sulfide 1 x 10-6 2.9 42 13-0.1 We have yet to identify any of the outgassed species from resist that contribute significantly to optics contamination in current power witness-type tests! 5 November 2008 11
Comparison of Injected Test Conditions to Outgassing from Resist Wafers Benzene is one of the common resist outgassing components From our outgassing measurements, a typical resist outgasses 5 x 10 13 molecules/cm 2 of benzene In our 8-hour, 36 million pulse, 29 J/cm 2 exposure, we require 2.8 x 10 20 molecules of benzene in a chamber with a pumping speed of 300 liters/second to keep the pressure at 1 x 10-6 Torr Equivalent to Benzene outgas from 8000 resist wafers sitting in the chamber during the testing Some outgas species are clearly not contributing to carbon contamination in witness-type testing at these power levels! Only data presented to date showing significant witness plate contamination by resist outgas species is NIST data [8] using high concentrations and 2,000 J/cm 2 dose. 5 November 2008 12
Can Anything Contribute Significant Contamination to Moderate Witness Plate Tests? Materials Tested with Witness Plate Apiezon vacuum grease Heated Apiezon vacuum grease Neoprene Heated neoprene Room temperature carbon conductive tape Heated (~ 100 s of degrees C) carbon conductive tape Filmetrix carbon thickness measured No signal above background No signal above background No signal above background No signal above background No signal above background 2 to 3 nm per exposure hour Tests done with EUV and broadband radiation (more likely to contaminate) total power density 0.1 to 1.0mW/cm 2. Only 1 material (heated Carbon tape) found to contribute witness plate contamination consistent with failure on ASML type test. Witness plate testing at these power levels does not show any problems with most organics. 5 November 2008 13
Discussion Current Scanners No data exists to date implicating resist outgassing in any tool optical contamination No data exists showing resist contamination significantly contributes to witness plate contamination above residual chamber organic contributions at current small-field and full-field power levels All tests to date indicate residual hydrocarbons are the dominant source of contamination at current power levels No justification has been shown for continuing resist outgas testing of any kind for small-field and current full-field low power scanners Significant resources are required to judge which resists should be tested, to arrange for testing, to perform the tests, and to report the results SEMATECH is discontinuing resist outgas testing for our MET tools with conventional PAG resists. 5 November 2008 14
Discussion Future Scanners Future scanners will need to improve orders of magnitude in residual organics before resist outgassing should be considered as any significant threat. Future scanners will hopefully have in situ carbon cleaning techniques; the presence of such capability should be fully comprehended in any budget analysis of allowable hydrocarbons from all sources. If in situ cleaning is successful on future scanners, then supercleanliness is not required, and resist outgassing is not of concern. If we need super-clean scanners, then developing techniques to qualify resists will be very difficult, as they need to meet the following requirements: Cheap Fast Accurate 5 November 2008 15
REFERENCES 1. L. Klebanoff, et al., Environmental data from the engineering test stand," Proc. SPIE, Vol. 4688, 310 (2002). 2. H. Cao, et al., Quantification of EUV Resist Outgassing," SEMATECH 3rd International EUVL Symposium, Poster session, November 2004. 3. 2004 SEMATECH Resist Advisory Group meeting 4. T. Aoki, K. Murakami, Acceptable Photoresist Outgassing," IEUVI Resist TWG meeting, October 2006. 5. B. Wolschrijn et al., New Method for Resist Outgas Qualification," IEUVI Resist TWG meeting, October 2006. 6. A. Wuest, Optics and Mask Contamination in SEMATECH EUV Micro-Exposure Tools," IEUVI Optics Contamination and Lifetime TWG meeting, March 2007 7. M. Chandhok, Intel Update Contamination of EUV MET mirrors," IEUVI Optics Contamination and Lifetime TWG meeting, March 2007 8. S. Hill et al., EUV Resist Outgassing: Data from NIST," IEUVI Optics Contamination TWG meeting, February 2008. 5 November 2008 16